Edition 360 | July 27, 2026 The Dyslexic AI Newsletter by LM Lab AI

Reading Time: 7 minutes Listening Time: 9 minutes

This weekend I left you with a question.

If AI is solving language, why did language need solving in the first place?

Here is the answer I have landed on, and it changes how I think about everything that comes after it.

Language was never natural. It was invented.

We treat language like it is as basic as breathing. It is not.

Somewhere in deep human history, a group of people needed to move a thought out of one skull and into another. Not a feeling. Not an instinct. A specific, structured piece of information. Danger is that direction. Food is over there. This plant will kill you.

Thought, on its own, cannot do that. Thought is trapped. It lives inside one brain and stays there unless something carries it out.

So humans built something. A system of sounds, then symbols, then rules for how to string them together, that could carry a thought out of one mind and rebuild something close to it inside another.

That system is language. And it is a technology, in the exact same sense that a wheel is a technology, or a printing press, or a telephone line.

We stopped noticing it as an invention a long time ago, the same way you stop noticing a wheel is technology once every car has four of them. But it was designed. It has a structure. And like every technology humans have ever built, that structure came with tradeoffs baked into it from the start.

Here is the tradeoff nobody talks about.

Language, as it was built, runs in a straight line. One word after another. One sentence after another. A beginning, then a middle, then an end.

Thought does not work that way.

Thought moves in clusters. In patterns. In sudden connections between two things that have no obvious relationship until you see them side by side. Thought can hold five things in tension at once. Language can only say one of them at a time.

That gap between how thought actually moves and how language forces it to move single file is not a flaw in you. It is a design constraint of the technology itself.

Every technology has one. A wheel is extraordinary at rolling forward and useless at climbing stairs. That is not a failure of the wheel. It is the shape of what it was built to do, and the shape of what it was never built to do.

Language was built to move information from one mind to another, one piece at a time, in order. It does that well. It was never built to preserve the shape of non-linear thought. It was never designed for that job.

This is the part that matters most to me personally.

If thought is naturally non-linear, and language is a technology that forces it into a straight line, then a mind that resists that forcing is not broken. It is running into the actual design limits of the tool.

I have spent a lot of editions of this newsletter describing the feeling of knowing something clearly and struggling to get it into words. I used to describe that as my problem. Something wrong with how I process language.

I don't think that anymore.

I think it is friction between a non-linear way of thinking and a linear technology that was never built to carry it.

That reframe matters. A wheel is not a bad wheel because it cannot climb stairs. You just would not use a wheel for that job. You would build a different tool, a set of legs, a different kind of technology entirely, if climbing stairs was what you actually needed to do.

Dyslexic and neurodivergent minds have been handed one tool, language in its narrowest, most linear form, and told that if we cannot use it fluently, something is wrong with us. Nothing is wrong with the wheel. And nothing is wrong with a mind that was built for a different kind of movement.

Here is a framework I want to name properly, because I am going to keep coming back to it.

I call it the Translation Stack.

Thought → Language → Mathematics → Computation → Language → Thought

Walk through it with me.

It starts as thought. Non-linear, clustered, alive in one mind.

It gets forced into language. Words, in order, one after another, the technology built to carry it out of your head.

When that language reaches an AI model, it becomes mathematics. Every word gets converted into numbers, vectors, coordinates in a mathematical space the model can actually operate on.

The model performs computation on that mathematics. Pattern matching, prediction, transformation, all of it happening in a layer that is not language at all anymore. It is pure math.

Then it comes back out as language again. Words, in order, reconstructed from the other side of the computation.

And finally, when another person reads or hears it, it becomes thought again. Non-linear, clustered, alive in a different mind now.

Thought to language to mathematics to computation to language to thought. Six steps. Two of them are the parts we actually notice, the language going in and the language coming out. The other four happen underneath, invisible, doing the real work of carrying meaning across the gap between two minds.

Here is the thing I think most people are missing.

AI did not invent a new way to think. It sped up a stack that has existed the entire time humans have communicated with each other.

Before AI, the middle of that stack, language becoming something else and then becoming language again, happened slowly, and mostly through other humans. A translator. An editor. A patient friend willing to sit with your half-formed idea until it made sense out loud.

AI did not add new steps to the stack. It made the existing steps faster, more flexible, and available to anyone with a phone, at any hour, without needing another human on the other end of the effort.

That is not a small thing. That is the entire reason this feels different.

Once you see the full stack, something shifts.

Language stops looking like the whole machine and starts looking like one interchangeable layer inside it.

You do not need to be fluent in every layer. You do not need to be a gifted writer to have a gifted mind. You need one working entry point into the stack, speaking, typing badly, sketching, gesturing, rambling into a voice memo, and one working exit point on the other side that your audience can actually receive.

Everything in between, language into mathematics, mathematics into computation, computation back into language, can now be handled by something other than your own effort alone.

That is the real shift. Not that AI writes for you. That the stack finally has enough flexible machinery in the middle that your entry point no longer has to match your exit point.

One thing to sit with this week.

Notice where your own entry point actually is.

Is it speaking out loud. Is it a messy bullet list. Is it a diagram you would draw on a napkin before you could ever write the paragraph that explains it. Whatever it is, that is your real entry point into the stack, whether or not it looks like "writing."

Stop trying to force your entry point to look like the exit point you eventually need. Let the stack do that work instead.

Use this the next time you catch yourself trying to think in finished sentences before you have even worked out the idea.

"I am going to give you my thinking exactly as it exists in my head right now: messy, out of order, possibly contradictory. Do not clean it up yet. First, just reflect it back to me in the same non-linear shape I gave it to you, so I can confirm you actually understood the pattern before we turn it into anything linear."

Confirm the pattern first. Worry about the sentence later.

Matt "Coach" Ivey Founder, LM Lab AI | Creator, The Dyslexic AI Newsletter

Dictated, not typed. Obviously.

🛠️ Language was never natural. It was invented, a technology built to move thought out of one mind and into another, the same way a wheel or a printing press is a technology.

📏 Every technology has a design and a cost. Language runs in a straight line. Thought does not. That mismatch is a design limit of the tool, not a flaw in your brain.

🧠 If you are dyslexic, this reframes everything. The friction you feel is not brokenness. It is a non-linear mind running into the limits of a linear technology that was never built to carry it.

🔁 New framework: the Translation Stack. Thought → Language → Mathematics → Computation → Language → Thought. Six steps. Only two of them are visible.

⚡ AI did not invent a new way to think. It sped up a stack that has existed the whole time humans have communicated. The middle layers used to require another human. Now they don't.

🔓 Language is one interchangeable layer, not the whole machine. Your entry point (speaking, sketching, rambling) no longer has to match your exit point. The stack in between can carry the difference.

🔒 Next edition keeps pulling on this thread, further underneath language, toward mathematics and pattern itself.

Edition 361: If language is a technology built on top of something else, what is actually underneath it? Edition 361 is called "The World Beneath Language," and it descends past language into pattern itself, the layer that language, mathematics, physics, and music are all secretly built out of.